Low-code AutoML-augmented Data Pipeline – A Review and Experiments. Issue 1 (February 2021)

• Record Type: Journal Article
• Authors: Gain, Ulla; Hotti, Virpi
• Is Part Of: Journal of physics. Volume 1828:Issue 1(2021)
• Publication Date: 2021-02
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/1828/1/012015
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: There is a lack of knowledge concerning the low-code autoML (automated machine learning) frameworks that can be used to enrich data for several purposes concerning either data engineering or software engineering. In this paper, 34 autoML frameworks have been reviewed based on the latest commits and augmentation properties of their GitHub content. The PyCaret framework was the result of the review due to requirements concerning adaptability by Google Colaboratory (Colab) and the BI (business intelligence) tool. Finally, the low-code autoML-augmented data pipeline from raw data to dashboards and low-code apps has been drawn based on the experiments concerned classifications of the "Census Income" dataset. The constructed pipeline preferred the same data to be a ground for different reports, dashboards, and applications. However, the constructed low-code autoML-augmented data pipeline contains changeable building blocks such as libraries and visualisations.